Summary Little is known about dementia disparities for Mexican Americans, who comprise the majority of United States (US) Latinos. Mexican Americans experience substantial social disadvantages compared with non-Latino whites in the US, including lower education, lower socioeconomic status, and migration-related stressors. Despite sustained disadvantage, the limited available evidence suggests that Mexican Americans exhibit similar rates of dementia as non-Latino whites. Unexpected health advantages of Mexican Americans have also been observed for many other health outcomes. Selective migration, both of healthy people from Mexico to the US and return migration of unhealthy adults back to Mexico, is a leading explanation for the Mexican American health advantage. Several known risk factors for dementia are also predictors of migration from Mexico to the US. However, no work has quantified selective migration and the bias it introduces to the comparison between Mexican American and non-Latino whites with respect to dementia risk. Therefore, it is unknown whether a dementia risk disparity exists for Mexican Americans after correcting for selective migration. In this study, we propose to combine data from two harmonized, nationally representative cohorts of adults aged 50+: the US Health and Retirement Study (HRS, n=26,690, including 2,452 Mexican Americans) and the Mexican Health and Aging Study (MHAS, n=13,798). These cohorts include harmonized longitudinal cognitive measures that are robustly linked to dementia risk. Accounting for selective migration is almost impossible with US-only data. Combining data from a Mexico- and a US- based cohort, we include both migrants and non-migrants, allowing us to test whether there are disparities in dementia risk and major factors driving dementia risk in Mexican Americans after accounting for migration patterns. This binational study will allow us to (Aim 1) quantify selective migration from Mexico to the US and return migration to Mexico as a function of established childhood and adult dementia risk factors that may influence migration patterns. We will then create a ?propensity to migrate? score. This score will adopt a new approach to evaluating the contribution of social factors to selective migration bias, based on evaluating relative social position within Mexican Americans born in Mexico, US-born Mexican Americans, and US-born non-Latino whites, rather than directly comparing people with the same absolute social position. Using this score to account for selection bias, we will show the magnitude of (Aim 2) differences in dementia risk comparing Mexican Americans to US non-Latino whites and (Aim 3) differences in dementia risk associated with migration to the US, compared to people who remained in Mexico. The link between socioeconomic factors and health in Mexican Americans is uncertain, so we will (Aim 4) identify effects of education, household and neighborhood disadvantage, and discrimination on dementia risk of Mexican Americans, accounting for selective migration. This study will provide critical information about dementia risk in this growing US population and other disadvantaged or immigrant groups.